423A35C7/LLaMA-Factory
1
0
Fork 0
mirror of https://github.com/hiyouga/LLaMA-Factory.git synced 2025-08-05 05:02:50 +08:00
Code Issues Packages Projects Releases Wiki Activity

Update trainer.py

Browse Source 
Former-commit-id: 6700a1b9fa0cbd965ac45d3f2de1088727235c25
This commit is contained in:
hoshi-hiyouga 2024-04-16 17:29:52 +08:00 committed by GitHub
parent b92f690190
commit 5226c4fa97
1 changed files with 3 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
src/llmtuner/train/sft/trainer.py
View File

@ -1,5 +1,6 @@
import json import json
import os import os
from types import MethodType
from typing import TYPE_CHECKING, Any, Dict, List, Optional, Tuple, Union from typing import TYPE_CHECKING, Any, Dict, List, Optional, Tuple, Union
import numpy as np import numpy as np
@ -9,8 +10,7 @@ from transformers import Seq2SeqTrainer
from ...extras.constants import IGNORE_INDEX from ...extras.constants import IGNORE_INDEX
from ...extras.logging import get_logger from ...extras.logging import get_logger
from ..utils import create_custom_optimzer, create_custom_scheduler from ..utils import create_custom_optimzer, create_custom_scheduler
from types import MethodType
from packaging import version
if TYPE_CHECKING: if TYPE_CHECKING:
from transformers.trainer import PredictionOutput from transformers.trainer import PredictionOutput
@ -31,6 +31,7 @@ class CustomSeq2SeqTrainer(Seq2SeqTrainer):
self.finetuning_args = finetuning_args self.finetuning_args = finetuning_args
if finetuning_args.use_badam: if finetuning_args.use_badam:
from badam import clip_grad_norm_for_sparse_tensor from badam import clip_grad_norm_for_sparse_tensor
self.accelerator.clip_grad_norm_ = MethodType(clip_grad_norm_for_sparse_tensor, self.accelerator) self.accelerator.clip_grad_norm_ = MethodType(clip_grad_norm_for_sparse_tensor, self.accelerator)
def create_optimizer(self) -> "torch.optim.Optimizer": def create_optimizer(self) -> "torch.optim.Optimizer":